Bug which mistakenly accuses the writer for shouting and issues a warning

Today, while typing an answer on Math.SE, I needed to use the factorial expression. While typing in my answer, I received the following warning:

DON'T SHOUT! All-caps posts are considered rude and will quickly be deleted.

Surprisingly, there were no letters in my answer, yet it warned me about typing in capitals.

Discuss (if possible): the mechanism of the CAPS detection function.

Bug to be fixed: training the algorithm to read for real alphabets instead of the number of exclamation marks.

Further research (on Chemistry, Physics and primary Meta): it was found that none of them really respond to caps or the above expression. For that reason, I checked my bug again on Math.SE, and it worked again.

Exact code:

$100! - 101! + ... - 109! + 110!$


Reference:

The third reference linked is stuck at evaluation and the bug has not been fixed. As per a comment :
This is caused by a blacklist entry for pattern ^(?-i:\P{Ll}+){15,}$, which matches 15 or more consecutive non-lowercase characters. Too bad you can't just use ^(?-i:\p{Lu}+){15,}$, since spaces and punctuation should not be considered to break the consecutive count. – Chris Jester-Young

The above report is definitely not valid now, look below:

looking at the above, please note there is a small a. Now look at the below one:

Now since it looks like it must be networkwide, here is an example of meta-stackexchange:

Update: After a bit of research on the S.E. network, I could not find any documentation to the warning triggered by the system. So before flagging a [bug] I would like to discuss how is the feature implemented and why is it site specific, i.e. only implemented on MSE and not on the sites namely physics, chemistry, meta stackexchange. The purpose of asking why is it site specific is that as far as I know, the backend code which can trigger such features makes it apply to the whole network.

• Yeah...and it doesn't show the error if the post does include caps, exclamation points, rude words etc. etc. etc. – Ollie Sep 5 at 19:54
• You have to click out of the text field to see any warnings. I tried entering the same text as you (a string of uppercase A's) and it was detected by the system. – Sonic the Masked Werehog Sep 7 at 5:07
• @SonictheMaskedWerehog now see the update, I typed a small "a" ago at the end (seen at the screenshot) – Queenie Goldstein Sep 7 at 5:14
• @SonictheMaskedWerehog Now I saw your screenshot, I think the "Test" word possibly failed the bug....and...that mouse pointer is BIG – Queenie Goldstein Sep 7 at 5:21
• Cagliostro, I understood that you replied to the wrong person; from the context of your comment. I've tested this on meta.math.se and a few other sites which support MathJax, as far as I can tell it ONLY :) occurs on the main math.se site and does not occur on MSE (here). --- Please demonstrate to yourself and give us a link to at least one more site where this occurs. An error that occurs on only one site should be posted on their meta, and not the main meta.; it would be off-topic here without any additional sites. At least you're up > 250! – Rob Sep 7 at 6:24
• @Rob testing is in progress on other sites ... I asked this question here assuming it is a networkwide implementation but I am surprised that a site can implement it's own features!(or the other sites rejected the global implementation). I am aware that there are certain differences between real meta and math meta for the reputations but this capitalization looks like a feature that should be on all sites. Seeing the test results, i think probably it was a hurry to put it on primary meta but I'll also wait to see the developer team comment on site specification of this feature – Queenie Goldstein Sep 7 at 6:31
• @Rob I have temporarily remove the bug tag so that we can discuss on the implementation of features which are site specific and features which are network wide – Queenie Goldstein Sep 7 at 8:27
• Maybe the page is reading your mind and anticipating that you are about to start SHOUTING! – GhostCat Sep 7 at 10:52
• @Rob the additional tags were supplied by another user. Second...studying your comment, I think I should re-include the bug tag. I undid the bug tag considering this line of yours -->more exceptions and special circumstances, the more cost and time intensive _I'll reinclude if you suggest this again because in no way am happy with the implementation which gets falsly triggered and can be avoided using just 1 letter not in caps. – Queenie Goldstein Sep 7 at 12:17
• It is possible, and not unheard of, to have a [discussion] [bug] [feature-request] tagged question - but I think the "discussion" has been had, it should work properly (not fouled by a small 'a'), and it's a "bug" that the feature doesn't work correctly on Math.SE; you'd like to "feature-request" that this be implemented on English language sites. Let's leave it for a bit, and not have Edititous (too much editing). :) --- BTW: I've deleted some of my comments, they are still available to Devs & Mods. – Rob Sep 7 at 13:19
• @Ollie Yes, I tried to test on math meta but did not know about the sandbox....so when I flagged my question for closure it was deleted and then i was told about the sandbox...so I'll remove that comm. – Queenie Goldstein Sep 9 at 4:06
• @Cagliostro copying and pasting your exact code doesn't provide you with an error here on Meta, funnily. – Ollie Sep 17 at 17:06
• @Ollie yes that has been seen(the second reference and the last image and even some comments), the capital warning is very centric and probably coded specifically into one site and forgotten by the developer to remove/upgrade – Queenie Goldstein Sep 17 at 17:50